Tutorial for Creating Graphs in SPSS
The graphical examples in this appendix are created using R and ggplot2. The SPSS version of these charts will look slightly different in terms of styling, colors, and exact layout, but the core information and structure will be the same. The R examples serve as a reference for what the final visualization should communicate.
If you are using the Core Dataset CSV files:
core_session.csv)Repeat for core_trials.csv when you need trial-level plots.
You will get better default chart behavior if variables are set correctly.
group, time, sex_category):
sprint_20m_s, vo2_mlkgmin, peak_force_n):
Optional but helpful: - Data → Sort Cases… (use time and group when you want organized output) - Data → Select Cases… (if you need to filter to a single time point, like pre only)
Use this when the variable is categorical and you want to show how many cases fall in each category (example: number of participants in training vs control).
# Load necessary libraries
suppressMessages(library(ggplot2))
suppressMessages(library(dplyr))
# Create sample data (replace with your actual data)
set.seed(123)
data <- data.frame(
group = sample(c("Control", "Training"), 100, replace = TRUE)
)
# Create bar chart for counts
ggplot(data, aes(x = group)) +
geom_bar(fill = "gray80", color = "black") +
labs(title = "Bar Chart (Counts): Group Distribution",
x = "Group",
y = "Count") +
theme_minimal()
Use this when you want relative frequencies (example: percent in each group, or percent injury yes vs no).
# Create bar chart for percentages
ggplot(data, aes(x = group)) +
geom_bar(aes(y = after_stat(count)/sum(after_stat(count))), fill = "gray70", color = "black") +
scale_y_continuous(labels = scales::percent) +
labs(title = "Bar Chart (Percentages): Group Distribution",
x = "Group",
y = "Percentage") +
theme_minimal()
Use this to compare category distributions across groups (example: sex category by group, injury status by group).
# Create sample data with sex_category
set.seed(123)
data_clustered <- data.frame(
group = sample(c("Control", "Training"), 200, replace = TRUE),
sex_category = sample(c("Male", "Female"), 200, replace = TRUE)
)
# Create clustered bar chart
ggplot(data_clustered, aes(x = sex_category, fill = group)) +
geom_bar(position = "dodge", color = "black") +
scale_fill_manual(values = c("Control" = "gray80", "Training" = "gray50")) +
labs(title = "Clustered Bar Chart: Sex Category by Group",
x = "Sex Category",
y = "Count",
fill = "Group") +
theme_minimal()
Use this to understand distribution shape for a quantitative variable (example: vo2_mlkgmin, sprint_20m_s, emg_rms_uv, sway_area_cm2).
sprint_20m_s)# Create sample quantitative data
set.seed(123)
data_hist <- data.frame(
sprint_20m_s = rnorm(100, mean = 3.5, sd = 0.3)
)
# Create histogram with properly scaled density curve
ggplot(data_hist, aes(x = sprint_20m_s)) +
geom_histogram(binwidth = 0.1, fill = "gray70", color = "black", alpha = 0.7) +
geom_density(aes(y = after_stat(density) * 0.1 * 100), alpha = 0.3, fill = "gray40", color = "black") +
labs(title = "Histogram: Sprint 20m Time",
x = "Sprint 20m (s)",
y = "Frequency") +
theme_minimal()
Use this to compare distributions or to see outliers (example: sprint time by group, sway area by time point).
# Create sample data for boxplot
set.seed(123)
data_box <- data.frame(
group = sample(c("Control", "Training"), 100, replace = TRUE),
sprint_20m_s = rnorm(100, mean = 3.5, sd = 0.3)
)
# Create boxplot
ggplot(data_box, aes(x = group, y = sprint_20m_s)) +
geom_boxplot(fill = "gray80", color = "black") +
labs(title = "Boxplot: Sprint 20m by Group",
x = "Group",
y = "Sprint 20m (s)") +
theme_minimal()
Use this when sample sizes are small or moderate and you want to show every observation (example: sprint time by group at pre).
SPSS dot plots are available through Legacy Dialogs and through some Chart Builder templates.
# Create dot plot
ggplot(data_box, aes(x = group, y = sprint_20m_s)) +
geom_dotplot(binaxis = "y", stackdir = "center", fill = "gray60", color = "black", binwidth = 0.05) +
labs(title = "Dot Plot: Sprint 20m by Group at Pre",
x = "Group",
y = "Sprint 20m (s)") +
theme_minimal()
Use this to visualize the relationship between two quantitative variables (example: peak_force_n vs sprint_20m_s, vo2_mlkgmin vs agility_ttest_s).
# Create sample data for scatterplot
set.seed(123)
data_scatter <- data.frame(
peak_force_n = rnorm(50, mean = 1000, sd = 200),
sprint_20m_s = rnorm(50, mean = 3.5, sd = 0.3)
)
# Create scatterplot
ggplot(data_scatter, aes(x = peak_force_n, y = sprint_20m_s)) +
geom_point(color = "black") +
geom_smooth(method = "lm", formula = y ~ x, se = FALSE, color = "black") +
labs(title = "Scatterplot: Peak Force vs Sprint 20m",
x = "Peak Force (N)",
y = "Sprint 20m (s)") +
theme_minimal()
Use this when group membership may explain clustering (example: training vs control).
group into “Set color” (if available in your chart template)group to panel rows or columns# Add group to scatter data
data_scatter_group <- data_scatter
data_scatter_group$group <- sample(c("Control", "Training"), 50, replace = TRUE)
# Create scatterplot with groups
ggplot(data_scatter_group, aes(x = peak_force_n, y = sprint_20m_s, color = group, shape = group)) +
geom_point() +
geom_smooth(method = "lm", formula = y ~ x, se = FALSE) +
scale_color_manual(values = c("Control" = "gray60", "Training" = "black")) +
scale_shape_manual(values = c("Control" = 16, "Training" = 17)) +
labs(title = "Scatterplot by Group: Peak Force vs Sprint 20m",
x = "Peak Force (N)",
y = "Sprint 20m (s)",
color = "Group",
shape = "Group") +
theme_minimal()
Use this to show average change across time points (pre, mid, post) for a quantitative variable (example: mean sprint time by group across time).
# Create sample data for line graph
set.seed(123)
data_line <- data.frame(
time = rep(c("Pre", "Mid", "Post"), each = 60),
group = rep(c("Control", "Training"), each = 30, times = 3),
sprint_20m_s = c(rnorm(30, 3.6, 0.3), rnorm(30, 3.4, 0.3), rnorm(30, 3.5, 0.3),
rnorm(30, 3.6, 0.3), rnorm(30, 3.3, 0.3), rnorm(30, 3.2, 0.3))
)
# Calculate means
data_line_summary <- data_line %>%
group_by(time, group) %>%
summarise(mean_sprint = mean(sprint_20m_s), .groups = 'drop')
# Create line graph
ggplot(data_line_summary, aes(x = time, y = mean_sprint, color = group, group = group, shape = group)) +
geom_line(linewidth = 1) +
geom_point(size = 3) +
scale_color_manual(values = c("Control" = "gray60", "Training" = "black")) +
scale_shape_manual(values = c("Control" = 16, "Training" = 17)) +
labs(title = "Line Graph: Mean Sprint 20m across Time by Group",
x = "Time",
y = "Mean Sprint 20m (s)",
color = "Group",
shape = "Group") +
theme_minimal()
Use this to show each participant’s change across two time points (example: function score pre vs post). This is often called a paired plot or slopegraph.
SPSS can do this using Legacy Line with “Values of individual cases” after restructuring or filtering to two time points.
Create a dataset that includes only pre and post rows for one outcome, and sort by id and time.
time = "pre" OR time = "post"id then time# Create sample data for paired plot
set.seed(123)
data_paired <- data.frame(
id = rep(1:20, each = 2),
time = rep(c("Pre", "Post"), 20),
function_0_100 = c(rnorm(20, 70, 10), rnorm(20, 75, 10))
)
# Create paired line plot
ggplot(data_paired, aes(x = time, y = function_0_100, group = id)) +
geom_line(alpha = 0.5) +
geom_point() +
labs(title = "Paired Line Plot: Function Score Pre to Post",
x = "Time",
y = "Function Score (0-100)") +
theme_minimal()
Use this to show each participant’s trajectory across multiple time points.
This is most useful for moderate sample sizes or when you split by group. If you include everyone in one plot, it can become visually dense.
# Create sample data for line graph
set.seed(123)
data_line <- data.frame(
id = rep(1:20, each = 9), # 20 participants, 3 time points, 3 groups? Wait, adjust
time = rep(rep(c("Pre", "Mid", "Post"), each = 20), 3), # Wait, better structure
group = rep(c("Control", "Training"), each = 30),
sprint_20m_s = rnorm(60, mean = 3.5, sd = 0.3)
)
# Actually, let's restructure properly
data_line <- expand.grid(id = 1:20, time = c("Pre", "Mid", "Post"), group = c("Control", "Training"))
data_line$sprint_20m_s <- rnorm(nrow(data_line), mean = 3.5, sd = 0.3)
# Create spaghetti plot
ggplot(data_line, aes(x = time, y = sprint_20m_s, group = id, color = group)) +
geom_line(alpha = 0.3) +
scale_color_manual(values = c("Control" = "gray60", "Training" = "black")) +
labs(title = "Spaghetti Plot: Sprint 20m across Time",
x = "Time",
y = "Sprint 20m (s)",
color = "Group") +
theme_minimal()
Use this to visualize patterns across trials (example: trial 1–3 jump height, peak force, sway area). This is typically done with core_trials.csv.
Work within a single time point first, then compare time points after you understand the trial pattern.
time = "pre"id then trial# Create sample trial data
set.seed(123)
data_trial <- data.frame(
id = rep(1:10, each = 3),
trial = rep(1:3, 10),
jump_height_cm = rnorm(30, mean = 40, sd = 5)
)
# Create trial-by-trial plot
ggplot(data_trial, aes(x = trial, y = jump_height_cm, group = id)) +
geom_line(alpha = 0.5) +
geom_point() +
labs(title = "Trial-by-Trial Plot: Jump Height at Pre",
x = "Trial",
y = "Jump Height (cm)") +
theme_minimal()
Use error bars to show uncertainty in group means across conditions or time points. Use these cautiously and pair them with raw points when sample sizes are not large.
# Create sample data for error bars
set.seed(123)
data_error <- data.frame(
group = sample(c("Control", "Training"), 50, replace = TRUE),
vo2_mlkgmin = rnorm(50, mean = 35, sd = 5)
)
# Calculate means and confidence intervals
data_error_summary <- data_error %>%
group_by(group) %>%
summarise(
mean_vo2 = mean(vo2_mlkgmin),
se = sd(vo2_mlkgmin) / sqrt(n()),
ci_lower = mean_vo2 - 1.96 * se,
ci_upper = mean_vo2 + 1.96 * se
)
# Create error bar plot
ggplot(data_error_summary, aes(x = group, y = mean_vo2)) +
geom_bar(stat = "identity", fill = "gray70", color = "black", alpha = 0.7) +
geom_errorbar(aes(ymin = ci_lower, ymax = ci_upper), width = 0.2, color = "black") +
labs(title = "Error Bars: VO2 by Group",
x = "Group",
y = "VO2 (mL/kg/min)") +
theme_minimal()
Use this to assess agreement between two measurement methods (example: comparing test-retest reliability, or different assessment tools for the same outcome).
SPSS 31 includes a dedicated Bland-Altman analysis feature:
method1_score and method2_score)This will automatically generate the Bland-Altman plot with:
Bland-Altman plots assess agreement between two measurement methods by plotting the difference between methods against their average. Here’s how to interpret the key elements:
Key Reference Lines:
Interpretation Guidelines:
Clinical/Practical Significance:
# Create sample data for Bland-Altman plot (paired measurements)
set.seed(123)
n <- 50
method1 <- rnorm(n, mean = 75, sd = 10)
# Method 2 has slight systematic bias (+2) and more variability
method2 <- method1 + 2 + rnorm(n, mean = 0, sd = 3)
# Calculate Bland-Altman metrics
mean_score <- (method1 + method2) / 2
difference <- method1 - method2
mean_diff <- mean(difference)
sd_diff <- sd(difference)
loa_upper <- mean_diff + 1.96 * sd_diff
loa_lower <- mean_diff - 1.96 * sd_diff
# Create Bland-Altman plot
bland_altman_data <- data.frame(mean_score, difference)
ggplot(bland_altman_data, aes(x = mean_score, y = difference)) +
geom_point(color = "black", alpha = 0.7) +
geom_hline(yintercept = mean_diff, color = "gray40", linetype = "solid", linewidth = 1) +
geom_hline(yintercept = loa_upper, color = "gray60", linetype = "dashed", linewidth = 0.8) +
geom_hline(yintercept = loa_lower, color = "gray60", linetype = "dashed", linewidth = 0.8) +
geom_hline(yintercept = 0, color = "black", linetype = "dotted", linewidth = 0.5) +
labs(title = "Bland-Altman Plot: Method Agreement",
x = "Mean of Two Methods",
y = "Difference (Method 1 - Method 2)") +
theme_minimal() +
annotate("text", x = min(mean_score) + 5, y = loa_upper + 1,
label = paste("Upper LoA:", round(loa_upper, 2)), hjust = 0, color = "gray40") +
annotate("text", x = min(mean_score) + 5, y = loa_lower - 1,
label = paste("Lower LoA:", round(loa_lower, 2)), hjust = 0, color = "gray40") +
annotate("text", x = min(mean_score) + 5, y = mean_diff + 1,
label = paste("Mean Diff:", round(mean_diff, 2)), hjust = 0, color = "gray40")
After a chart is created, you can edit labels, scales, and formatting in the Chart Editor.
Common edits:
If you want consistent image sizes, set output options before exporting:
For each analysis, save at least one plot that shows the raw data pattern and one plot that shows a summary. This prevents summary-only reporting from hiding important structure.